[Bugfix] Fix cache block size calculation for CPU MLA (vllm-project#15848)

Signed-off-by: Thien Tran <[email protected]>

Author: gau-nernst

vllm/worker/cpu_worker.py

@@ -111,7 +111,7 @@ def get_cache_block_size(
         num_layers = model_config.get_num_layers(parallel_config)
 
         key_cache_block = block_size * num_heads * head_size
-        value_cache_block = key_cache_block
+        value_cache_block = key_cache_block if not model_config.use_mla else 0
         total = num_layers * (key_cache_block + value_cache_block)
         if cache_dtype == "auto":
             dtype = model_config.dtype